Azure AI Apps and Agents Developer (AI-103) practice questions

Question 1

A Microsoft Foundry model generated short promotional product videos. Reviewers find a small watermark in the top-right corner of some approved clips. You must remove the watermark without regenerating the videos. What should you do?

Accepted Answer

Apply a mask-based inpainting edit to the affected part of the video.

Answer

Increase the guidance scale.

Answer

Crop the video by using the size parameter.

Answer

Modify the original prompt to exclude watermarks.

Question 2

A Microsoft Foundry customer-support agent must ground accurate, context-aware responses in internal documentation stored in Azure AI Search, perform deep multi-step reasoning across long contexts, and generate detailed natural-language responses. Which type of model should power the agent?

Accepted Answer

A large language model (LLM)

Answer

A text embedding model

Answer

A key phrase extraction model

Answer

A small language model (SLM)

Question 3

You have 5,000 images of scanned invoices in a file share. You must extract invoice items, sales amounts, and customer details. What should you use?

Accepted Answer

Azure Document Intelligence in Foundry Tools

Answer

Azure Custom Vision

Answer

Azure AI Immersive Reader

Answer

Azure Vision in Foundry Tools

Question 4

You are building an image-sharing app that uses Azure AI to prevent users from sharing sexually explicit images. You must identify inappropriate images accurately while minimizing development effort. What should you use?

Accepted Answer

Content Safety in Foundry Control Plane Studio

Answer

Vision Studio in Azure Vision in Foundry Tools

Answer

Visual Studio

Answer

Microsoft Foundry

Question 5

A GitHub repository contains a YAML file named File1 that defines the evaluation settings for a Microsoft Foundry agent. You need a GitHub Actions workflow to run the evaluation defined in File1 whenever a pull request opens. How should you configure the workflow?

Accepted Answer

Set eval-results-config to the path of File1.

Answer

Set tenant-id to the Microsoft Entra tenant ID.

Answer

Set project-endpoint to the endpoint of the project.

Answer

Set model-deployment-name to the deployed model.

Question 6

You successfully call the Microsoft.CognitiveServices accounts/regenerateKey REST operation with a request body of { "keyName": "Key2" }. What is the result?

Accepted Answer

A new secondary subscription key is created.

Answer

A new query key is generated.

Answer

The primary and secondary subscription keys are both rotated.

Answer

A key for Azure AI Services is generated in Azure Key Vault.

Question 7

A Microsoft Foundry agent uses Azure AI Search as its retriever. You will ingest PDFs containing text and embedded images, and an OCR skill must process the images during indexing. Which indexing approach should you use?

Accepted Answer

Use an indexer to extract image data into a normalized_images collection.

Answer

Use a skillset to run OCR directly against the index content field.

Answer

Use outputFieldMappings to write image data to a searchable field.

Answer

Use a Shaper skill to restructure the OCR input.

Question 8

A Microsoft Foundry application processes PDF vendor invoices. You need Azure Document Intelligence in Foundry Tools to generate Markdown output that preserves sections and table structure while minimizing development effort. What should you do?

Accepted Answer

Set outputcontentformat=ContentFormat.MARKDOWN.

Answer

Configure output figures when analyzing the PDF.

Answer

Configure content markdown when analyzing the document.

Answer

Increase the confidence threshold.

Question 9

A Foundry agent grounds responses in an Azure AI Search index with searchable product names and product codes plus vector embeddings for product descriptions. Users must query by exact names or codes and by natural-language descriptions. What should you configure?

Accepted Answer

Hybrid search

Answer

Vector search only

Answer

Semantic search only

Answer

Keyword search only

Question 10

A Foundry agent ingests scanned PDF vendor invoices with tables and embedded QR codes. You need Azure Content Understanding in Foundry Tools to preserve layout, extract content and layout elements, and detect QR codes without a language-model deployment. Which built-in analyzer should you use?

Accepted Answer

prebuilt-layout

Answer

prebuilt-read

Answer

prebuilt-documentSearch

Answer

prebuilt-documentFieldSchema

Azure AI Apps and Agents Developer (AI-103) practice questions

Question stream

Discussion

Manage your cookie preferences

Loading subject practice

Loading Brainliest

Loading subject practice

Azure AI Apps and Agents Developer (AI-103) practice questions

Question stream

Discussion

Practice overview